3D Facial Landmarks and Face-Mesh Generation via TensorFlow and ReactJS
https://meet.google.com/zxx-ycmy-qrs
Time: May 26th, 2023, Friday, 18:30
Lecturer: Atef Alreni
Meeting: https://meet.google.com/zxx-ycmy-qrs
3D facial landmark detection is essential for face recognition, face tracking, AR filters such as those used in Meta’s Facebook and Instagram, et cetera. The aforementioned are sophisticated technological capabilities that supply numerous real-world applications, e.g., Identity verification, border control and Immigration and law enforcement and security. However, visually comprehending 3D landmarks and their association with one another can be perplexing. Thus, this project aims to generate and separately visualise 468 3D facial landmarks and a face-mesh via TensorFlow’s pretrained face-mesh model and ReactJS. The project allows users to visually discern how these landmarks are connected and how they unite to produce a face mesh. The project further will allow users to upload an image with faces that will result in three images: the original image, an image with 468 landmarks on all faces and a final image with a face mesh on all faces. The project will leverage the capabilities of TensorFlow and ReactJS to provide an intuitive app that supplies an understanding of 3D landmark detection and the association of landmarks.
This event is part of the IAPS series on Machine Learning and is organized jointly by the Institute for Advanced Physical Studies, Quanterall and Varna Free University.
